Flowchart identifying individuals with rheumatoid arthritis (RA) and inflammatory bowel disease (IBD) who met the inclusion and exclusion criteria for study. Other dermatologic conditions consisted of xeroderma pigmentosa and albinism. Other immunosuppressive therapies included tacrolimus, cyclosporine, imiquimod, or fluorouracil. Cumulative exposure time is reported in median (interquartile range) and is measured among those who had received the drug. HIV indicates human immunodeficiency virus; NMSC, nonmelanoma skin cancer.
eTable 1. Diagnostic and Procedural Codes Used for Identification of NMSC
eTable 2. Cumulative Exposure to Medication in Those Who Developed Second NMSC vs Those Who Did Not
eTable 3. Impact of Duration of Exposure to Anti-TNFs and Thiopurines in RA and IBD on Risk of Second NMSC
eFigure. Schematic of Criteria for Cohort Entry
Customize your JAMA Network experience by selecting one or more topics from the list below.
Scott FI, Mamtani R, Brensinger CM, et al. Risk of Nonmelanoma Skin Cancer Associated With the Use of Immunosuppressant and Biologic Agents in Patients With a History of Autoimmune Disease and Nonmelanoma Skin Cancer. JAMA Dermatol. 2016;152(2):164–172. doi:10.1001/jamadermatol.2015.3029
Immune dysfunction underlies the pathogenesis of rheumatoid arthritis (RA) and inflammatory bowel disease (IBD). Immunosuppressive therapy is the standard of care for these diseases. Both immune dysfunction and therapy-related immunosuppression can inhibit cancer-related immune surveillance in this population. Drug-induced immunosuppression is a risk factor for nonmelanoma skin cancer (NMSC), particularly squamous cell tumors. For patients with a history of NMSC, data are limited on the effect of these drugs on the risk of additional NMSCs.
To determine the relative hazard of a second NMSC in patients with RA or IBD who use methotrexate, anti–tumor necrosis factor (anti-TNF) therapy, or thiopurines after an initial NMSC.
Design, Setting, and Participants
In this retrospective cohort study, we studied 9460 individuals with RA or IBD enrolled in Medicare from January 1, 2006, through December 31, 2012.
Exposure to methotrexate, thiopurines, anti-TNFs, sulfasalazine, hydroxychloroquine, abatacept, or rituximab after the incident NMSC surgery.
Main Outcomes and Measures
A second NMSC occurring 1 year or more after the incident NMSC using Cox proportional hazards regression models.
Among 9460 individuals (6841 with RA and 2788 with IBD), the incidence rate of a second NMSC per 1000 person-years was 58.2 (95% CI, 54.5-62.1) and 58.9 (95% CI, 53.2-65.2) in patients with RA and IBD, respectively. Among patients with RA, methotrexate used in conjunction with other medications was associated with an increased risk of a second NMSC (hazard ratio [HR], 1.60; 95% CI, 1.08-2.37). Adjusted for other medications, the risk of NMSC increased with 1 year or more of methotrexate use (HR, 1.24; 95% CI, 1.04-1.48). Compared with methotrexate alone, the addition of anti-TNF drugs was significantly associated with risk of NMSC (HR, 1.49; 95% CI, 1.03-2.16). Abatacept and rituximab were not associated with increased NMSC risk. The nonsignificant HRs for 1 year or more of thiopurine and anti-TNF use for IBD were 1.49 (95% CI, 0.98-2.27) and 1.36 (95% CI, 0.76-2.44), respectively.
Conclusions and Relevance
Methotrexate use is associated with an increased risk of a second NMSC. Anti-TNF use may increase the risk of a second NMSC when used with methotrexate for RA. Further long-term studies are required before one can conclude that thiopurine and/or anti-TNF do not increase the risk of a second NMSC in patients with IBD.
The incidence of nonmelanoma skin cancer (NMSC) is increasing in white populations worldwide.1 Dermatologists, rheumatologists, gastroenterologists, and primary care physicians are capable of recognizing NMSC lesions, with most being treated by dermatologists.2 Although initial surgical treatment is usually curative, the risk of a second primary NMSC is high.3 Major risk factors for NMSC are skin pigmentation and solar damage to the skin.4 Medications that accelerate the phototoxic process have been associated with an increased incidence of NMSC.5 Several medications are considered photosensitizers, including methotrexate and thiopurines.6,7 This effect has not been reported with biologic therapies. Immunosuppression is also believed to be a risk factor for NMSC, particularly squamous cell carcinoma, which has prompted frequent screening of individuals after solid organ transplant.8
Immune dysfunction underlies the pathogenesis of rheumatoid arthritis (RA) and inflammatory bowel disease (IBD). Immunosuppressive therapies have become the standard for treating these diseases.9,10 Methotrexate was one of the first therapies with a demonstrated benefit in treating RA and has remained a cornerstone therapy.11 Similarly, in IBD, thiopurine use has increased notably in recent decades.12 These therapies have been further augmented by medications that target tumor necrosis factor α (anti-TNFs) and other targets, collectively referred to as biologic therapies.13,14 These agents are used alone or in combination with methotrexate or thiopurines.15-20
Methotrexate, thiopurines, and anti-TNFs have been associated with hematologic and dermatologic malignancy.21-23 Thiopurines appear to increase the risk of NMSC during active use of the drug and possibly after use of the medication is discontinued.24,25 Similar effects were noted with methotrexate in a systematic review that assessed the risk of NMSC in patients with psoriasis.26,27 The association between anti-TNFs and NMSC is less clear, with conflicting results among several studies.22,27,28
For patients with a prior malignancy, there are limited data regarding the effect of these medications on the risk of cancer recurrence or a second primary tumor. The available data are generally in small cohorts and have combined different cancer types.29-31 In this study, we assessed the risk of a second NMSC in Medicare beneficiaries with RA or IBD exposed to methotrexate, thiopurines, or biologic agents. We hypothesized an increased risk of a second NMSC among patients with exposure to photosensitizers, such as methotrexate or thiopurines, relative to those who have received biologic agents without these effects, such as anti-TNFs or nonimmunosuppressive therapies.
We performed a retrospective cohort study among patients with RA or IBD using Medicare data from January 1, 2006, through December 31, 2012. This cohort has been used previously to evaluate the comparative effectiveness and safety of medications used to treat IBD and RA.28,32-34 The study protocol was approved by the University of Alabama at Birmingham and University of Pennsylvania institutional review boards. No informed consent was required.
Individuals 18 years or older with a diagnosis of RA or IBD based on International Classification of Diseases, Ninth Revision diagnosis codes and an incident NMSC diagnosis after enrollment in Medicare from January 1, 2006, through December 31, 2010, were considered eligible for this study. Follow-up time continued for this cohort until December 31, 2012. Incident diagnoses of NMSC were identified using an adapted claims-based algorithm,35 combining diagnostic codes for NMSC and dermatologic procedures (eMethods in the Supplement).
Study participants were required to have a baseline observation period of 6 months before the first NMSC diagnosis to assess medication exposures before or at the time of NMSC diagnosis.36 In addition, patients were required to have 12 months or more of follow-up time after the first NMSC diagnosis or procedure, without an additional NMSC diagnostic or procedure code from 6 to 12 months to maximize the likelihood that subsequent NMSC codes represented incident events instead of follow-up for a prevalent NMSC (eFigure 1 in the Supplement). This follow-up window allowed 1 year for the completion of therapy for the incident NMSC event before diagnosis of a second NMSC; similar methods have been used with administrative data to capture recurrent and second malignancies.37
Individuals were excluded if they had an NMSC diagnosis within the first 6 months of enrollment or diagnosis with any of the following conditions before the first NMSC diagnosis: any malignancy, psoriasis, organ transplant, human immunodeficiency virus, xeroderma pigmentosa, or albinism.38,39 We excluded patients with any recorded use of medications thought to affect the risk of NMSC before the first NMSC diagnosis, such as tacrolimus, cyclosporine, imiquimod, or fluorouracil. Individuals enrolled in Medicare Part C (Managed Medicare) were excluded because they may have incomplete drug and outcome data. Follow-up was censored if patients met any exclusion criteria after the first NMSC diagnosis.
The following medications were considered exposures of interest: (1) methotrexate, (2) thiopurines (azathioprine or 6-mercaptopurine), (3) anti-TNFs (infliximab, adalimumab, certolizumab, golimumab, or etanercept), (4) leflunomide, (5) tocilizumab, (6) abatacept, (7) rituximab, and (8) sulfasalazine or hydroxychloroquine. Exposure was defined as 2 or more dispensings or infusions within 4 months of each other, with at least 1 dispensing or infusion after the incident NMSC surgery. Exposure was further divided into current and recent, where recent exposure began 90 days after the expected end of each prescription or infusion dosing interval if use of the medication was not continued. In addition, time updating variables that described cumulative exposure were generated for each medication. We assessed the duration of exposure for methotrexate, anti-TNFs, and thiopurines, categorized as never exposed and less than 1 year, 1 to 2 years, 2 to 3 years, and more than 3 years of exposure.
Follow-up began 1 year after the first NMSC surgery. Follow-up ended with the earliest of the following: (1) subsequent new NMSC diagnosis using the same criteria as described above, (2) death, (3) loss of medical or prescription benefits, or (4) end of data collection.
Potential confounders were assessed at the start of follow-up, including age (in deciles), sex, race/ethnicity, urban vs rural residence, Charlson Comorbidity Index scores, and nursing home inhabitance. Latitude was stratified as greater than or equal to or less than the median for the cohort. A history of actinic keratosis and the number of dermatologic visits in the 12-month period between NMSC diagnosis and start of follow-up were assessed. Exposure to the medications of interest before the initial NMSC was evaluated as a potential confounder using all available data before the start of follow-up. Cumulative corticosteroid exposure was assessed as a confounder as a time-varying covariate measured in prednisone-equivalent dosing and categorized as no exposure, less than 1.5 g, and 1.5 g or greater.
Analyses were performed using SAS statistical software, version 9.3 (SAS Institute Inc), and STATA software, version 13.0 (StataCorp). Analyses were performed separately for patients with RA and IBD. Our data use agreement with the Centers for Medicare & Medicaid Services precluded reporting results in which fewer than 10 patients were exposed to a therapy or experienced the outcome. Baseline covariates were assessed using χ2 and Fisher exact tests as appropriate. We computed separate disease-specific Cox proportional hazards regression models for each exposure contrast of interest, adjusted for covariates. Covariates were selected for final multivariable models if inclusion changed the hazard ratio (HR) for the primary exposure by 10% or more. We computed time-updating HRs of each exposure of interest, adjusting for potential confounders. A test for interaction between our exposures of interest and follow-up time was used to ensure that proportional hazards assumptions were not violated.
Because methotrexate is commonly used in combination with other therapies (hydroxychloroquine, sulfasalazine, or anti-TNF), we assessed the risk of NMSC with methotrexate plus hydroxychloroquine or sulfasalazine vs hydroxychloroquine or sulfasalazine alone and methotrexate plus anti-TNF vs anti-TNF alone. These analyses were limited to patients with documented methotrexate use before the first NMSC diagnosis. We computed the pooled association of methotrexate vs no methotrexate with a second NMSC using fixed-effects meta-analysis.40,41 Similar analyses were conducted for thiopurines and anti-TNFs in the IBD cohort.
There were 9460 individuals within the cohort: 6672 with RA only, 2619 with IBD only, and 169 with both RA and IBD (Figure). A total of 1291 individuals developed a second NMSC (910 with RA only, 359 with IBD only, and 22 with both RA and IBD). The incidence rate per 1000 person-years of a second NMSC was 58.2 (95% CI, 54.5-62.1) and 58.9 (95% CI, 53.2-65.2) among patients with RA and IBD, respectively. For both the patients with RA and those with IBD, men were more likely to have a second NMSC compared with women (P < .001 and P = .01, respectively) (Table 1). The median latitude for the cohort was 37.5°. Latitude was not associated with an increased risk of a second NMSC in patients with RA or IBD. There was a greater prevalence of a history of actinic keratosis in those with a second NMSC in both the RA and IBD groups (P < .001).
The median methotrexate exposure time after an initial NMSC in the patients with RA was 1.64 years (interquartile range, 1.15-2.31 years) and 2.61 years (interquartile range, 1.80-3.80 years) among those with and without a second NMSC, respectively. Among sulfasalazine or hydroxychloroquine users, there was an increased risk of a second NMSC with methotrexate exposure, although this finding was not statistically significant (HR, 1.81; 95% CI, 0.94-3.52) (Table 2). There was similarly an increased but not statistically significant risk of a second NMSC with methotrexate use in patients also treated with anti-TNFs (HR, 1.50; 95% CI, 0.92-2.44). When meta-analytic methods were used to pool these groups, methotrexate exposure was associated with an increased risk of a second NMSC (HR, 1.60; 95% CI, 1.08-2.37).
We evaluated the effect of duration of methotrexate therapy after incident NMSC, categorizing methotrexate exposure as never exposed, short-term exposure (<1 year), and exposure of 1 year or more. We observed an increased risk of NMSC with increasing duration of exposure, particularly among those exposed for 1 year or more (HR, 1.24; 95% CI, 1.04-1.48) (Table 3). After adjusting for age, sex, other immunosuppressive therapies, and latitude, the risk of a second NMSC increased with longer methotrexate exposure (<1 year: HR, 1.10; 95% CI, 0.84-1.44; 1-2 years: HR, 1.16; 95% CI, 0.95-1.41; 2-3 years: HR, 1.36; 95% CI, 1.05-1.77; and >3 years: HR, 1.59; 95% CI, 1.09-2.32).
Among patients treated with methotrexate for RA, there was no significantly increased risk with rituximab or abatacept use, although anti-TNF use was statistically significant when adjusted for anti-TNF use before the first NMSC (HR, 1.49; 95% CI, 1.03-2.16) (Table 2). When stratified by exposure duration, short-term anti-TNF use was significantly associated with increased risk (<1 year: HR, 1.43; 95% CI, 1.01-2.04), but longer use was not (Table 3 and eTable 3 in the Supplement). Risks attributable to leflunomide were not calculated because of limited numbers of events. Cumulative corticosteroid exposure was not significantly associated with risk of a second NMSC in RA (P = .53).
Among patients with IBD, thiopurine use was not associated with increased risk of a second NMSC when used in combination with anti-TNF vs anti-TNF monotherapy (HR, 0.79; 95% CI, 0.30-2.08) (Table 2). In comparisons of anti-TNF vs thiopurine monotherapy, thiopurines appeared to have a higher incidence rate of a second NMSC, although this was not statistically significant. Longer duration of anti-TNF use was not associated with an increased risk of a second NMSC (eTable 3 in the Supplement). The risk of a second NMSC was not significantly increased with short-term thiopurine therapy (HR, 1.53; 95% CI, 0.87-2.70) and was not significantly increased with 1 year or more of thiopurine therapy (HR, 1.49; 95% CI, 0.98-2.27) (Table 4). When further stratified by duration of exposure, the degree of risk remained similar although still not statistically significant (2-3 years: HR, 1.57; 95% CI, 0.83-2.97; >3 years: HR, 1.49; 95% CI, 0.60-3.73). Cumulative corticosteroid exposure was not significantly associated with a second NMSC in IBD (P = .89).
In this retrospective cohort study, we examined the effect of immunosuppressant therapies on the risk of a second NMSC in patients with RA and IBD. We hypothesized that immunosuppressant medications used to treat RA and IBD, particularly methotrexate and thiopurines, which are photosensitizing, may increase the risk of a second NMSC.5 Among individuals with RA, methotrexate and anti-TNF drugs were both associated with an increased risk of a second NMSC diagnosis. A similar association was seen for anti-TNFs among the IBD cohort, although this finding was not statistically significant. Abatacept or rituximab in conjunction with methotrexate were not associated with a significantly increased risk for a second NMSC compared with methotrexate monotherapy, although the point estimates were similar to anti-TNF agents and sample sizes were small. Consequently, one cannot interpret these results as indicating that these agents are safer alternatives. Among individuals with IBD, thiopurine use was not significant for the association with an increased risk of a second NMSC, and the estimated relative risk was similar to that seen for methotrexate in RA.
For patients with RA, the incidence of a second NMSC was increased by 19 per 1000 person-years when methotrexate was used with anti-TNFs and by 34 per 1000 person-years when methotrexate was used in addition to nonimmunosuppressive therapies. This increase in incidence rate translates to a number needed to treat to cause 1 additional NMSC per year of 52.6 and 29.4 when used with anti-TNFs or without concomitant immunosuppressant medications, respectively. Given that methotrexate is generally the first-line therapy for RA, with other drugs typically added to methotrexate when needed, these data emphasize the need for intensive NMSC surveillance protocols.
The evidence of an increased risk of a second NMSC among those treated with thiopurines for IBD was not statistically significant although consistent with several prior studies24,25 reporting that thiopurines increase the incidence of a first NMSC. This failure to achieve statistical significance may have been owing to inadequate statistical power or may reflect a persistent effect of prior thiopurine exposure, which would be expected among many of the anti-TNF–treated patients in this cohort.24 Likewise, if both thiopurines and anti-TNFs increase the risk of a second NMSC, the association would be attenuated in direct comparisons of these 2 drug classes.
Anti-TNF therapy was significantly associated with the risk of a second NMSC among patients with RA, and the magnitude of risk was comparable in patients with IBD. Prior studies22,27,28 in IBD have suggested an increased risk of melanoma with anti-TNF therapy, but there is less evidence for an increased risk of NMSC. In RA, the association between an initial NMSC and anti-TNFs appears more firmly established, although the association is often complicated by concomitant methotrexate use.42-46 One can hypothesize that immunosuppression from anti-TNFs might contribute to NMSC risk, particularly in patients with a prior NMSC. In contrast to thiopurines and methotrexate, anti-TNF therapy is not photosensitizing. Regardless, these data suggest that anti-TNF therapy may not be an advantageous option in place of thiopurines or methotrexate for patients with a prior NMSC history.
This study has several important strengths. Nonmelanoma skin cancer is the most frequently diagnosed malignant neoplasm in the United States.47 Therefore, the risk of recurrence is an important question commonly faced in clinical practice, and research examining this risk has been limited to case reports and series.29-31 This is the largest study to date, to our knowledge, to examine the effect of commonly used therapies in IBD and RA on the risk of cancer recurrence. Furthermore, Medicare coverage comprises a geographically diverse patient population and should be generalizable to approximately 93% of older adults in the United States.48
There are several potential limitations of this study. As with any retrospective study using claims-based data, there is the risk of misclassification. However, requiring both diagnostic and procedural codes to define NMSCs minimizes the probability of including individuals without a first NMSC in the cohort by maximizing the positive predictive value of a true incident event. There is also the potential for surveillance bias because individuals with a prior NMSC who receive methotrexate or anti-TNF drugs may be more frequently surveyed than individuals not receiving these medications. To account for this factor, we adjusted for dermatology visits in the first year after the first NMSC. We are unable to ensure that our diagnostic codes identified exclusively second NMSC events as opposed to the initial event. However, we used a 1-year window after the initial diagnosis to minimize this misclassification. Although we were unable to specifically measure disease severity in RA or IBD, we used corticosteroid exposure as a surrogate and found no significant association with a second NMSC in univariate analysis and only one model with evidence of confounding. Because we studied NMSC, we cannot generalize these results to the risk of recurrence of other cancers. Our analysis was restricted to a predominantly white population, although this is the population that is at greatest risk for NMSC. We were also unable to assess NMSC risk according to specific ethnicity or sunscreen use.
We had limited power to study some of the drugs of interest. There were too few patients exposed to leflunomide. Sample sizes were similarly small for analysis of abatacept, tocilizumab, and rituximab. The estimated HR with rituximab was among the highest observed, and there was some evidence of greater risk with longer therapy, but the CIs were wide. Because rituximab is recommended among the biologic therapies for use in patients with a prior history of cancer in the American College of Rheumatology 2012 guidelines,49 this estimate may be biased from confounding by indication. Likewise, we were also unable to assess the effect of new thiopurine use in patients who were thiopurine naive before their initial NMSC event because of limited numbers. This finding may represent reluctance on the part of physicians to initiate therapy with thiopurines in individuals who have a known NMSC. Therefore, although this study found an increased risk of a second NMSC with methotrexate and possibly anti-TNF drugs, we cannot conclude that the other immunosuppressive therapies represent safer alternatives. Further research examining these agents is required.
The use of immunosuppressive therapy is known to be a risk factor for NMSC, most commonly squamous cell carcinoma. Physicians treating patients with RA and IBD face the decision of what medications to use in patients with a history of NMSC. In this study, methotrexate use was associated with an increased risk of a second NMSC in RA, with increasing risk with longer duration of exposure. Although there appears to be an association between thiopurine use and an initial NMSC in IBD, we did not observe a statistically significant increased risk for a second NMSC, although this may have been due to reduced statistical power because the estimated relative risk with thiopurines for IBD was larger than that observed with methotrexate for RA. Last, anti-TNF therapy may further increase the risk of a second NMSC, particularly when used in conjunction with methotrexate to treat RA. These data can be used to guide therapeutic decisions in patients with prior NMSC.
Accepted for Publication: July 17, 2015.
Corresponding Author: Frank I. Scott, MD, MSCE, 3400 Civic Center Blvd, Seventh Floor, PCAM 7 SPE, Philadelphia, PA 19104 (firstname.lastname@example.org).
Published Online: October 28, 2015. doi:10.1001/jamadermatol.2015.3029.
Author Contributions: Drs Curtis and Lewis had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Scott and Mamtani contributed equally to this work and served as co-first authors. Drs Curtis and Lewis served as co-senior investigators and co-last authors.
Study concept and design: Scott, Mamtani, Zhang, Chen, Yun, Osterman, Beukelman, Margolis, Curtis, Lewis.
Acquisition, analysis, or interpretation of data: Scott, Mamtani, Brensinger, Haynes, Chiesa-Fuxench, Zhang, Xie, Yun, Osterman, Beukelman, Curtis, Lewis.
Drafting of the manuscript: Scott, Mamtani.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Scott, Mamtani, Brensinger, Chen, Xie, Beukelman, Lewis.
Obtained funding: Curtis, Lewis.
Administrative, technical, or material support: Haynes, Margolis, Curtis.
Study supervision: Curtis.
Conflict of Interest Disclosures: Dr Lewis reported serving as a consultant for Takeda, Amgen, Millennium Pharmaceuticals, Prometheus, Lilly, Shire, AstraZeneca, Janssen Pharmaceuticals, Merck, and AbbVie and on a data and safety monitoring board for clinical trials sponsored by Pfizer. He reported receiving research support from Bayer, Shire, Centocor, Nestlé, and Takeda. Dr Curtis reported serving as a consultant for Roche/Genentech, UCB, Janssen Pharmaceuticals, CORRONA, Amgen, Pfizer, BMS, Crescendo, and AbbVie and receiving research support from Roche/Genentech, UCB, Janssen Pharmaceuticals, CORRONA, Amgen, Pfizer, BMS, Crescendo, and AbbVie. Dr Beukelman reported serving as a consultant for Genentech, Novartis, and UCB and receiving research support from Pfizer. Dr Zhang reported receiving research support from Amgen. Dr Osterman reported serving as a consultant for Janssen Pharmaceuticals, Abbott, and UCB and receiving research support from UCB. Dr Haynes reported receiving research support from AstraZeneca/Bristol-Meyers Squibb. Dr Mamtani reported serving as a consultant for Takeda. No other disclosures were reported.
Funding/Support: This study was supported in part by grants R01 HS018517 and U19 HS021110 from the Agency for Healthcare Research & Quality (Dr Curtis) and grants K08-DK095951 (Dr Scott), K12 CA 076931 (Dr Mamtani), and K24-DK078228 (Dr Lewis) from the National Institutes of Health.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.